Ndea

Founded by François Chollet in 2025

Combine program synthesis and deep learning to advance artificial intelligence research

Pricing model

Demo

Rating

★ Mixed Reviews

Time to value

Requires Expertise

You can use Ndea to explore frontier artificial intelligence architectures designed for complex scientific discovery. The lab combines deep learning pattern recognition with guided program synthesis to create models capable of rigorous formal reasoning from minimal training data. Instead of merely interpolating known data points, these systems synthesize discrete programs that explain observations and generalize across novel scenarios. Researchers work with these architectures to tackle open-ended challenges across scientific fields such as drug discovery, robotics, sustainable energy systems, and autonomous space exploration.

What Ndea does

Synthesize computer programs from sparse observation dataSearch discrete model spaces guided by neural networksFormulate logical rules to explain complex datasetsGenerate verified algorithmic answers to open-ended problemsAcquire reusable abstract skills from minimal training examplesExecute autonomous reasoning experiments for scientific researchGuided program synthesis architectureDiscrete program search guided by deep learningIntegration of formal logical reasoning with pattern recognitionSample-efficient learning from minimal training examplesAutonomous abstraction and skill acquisition capabilitiesArchitectures designed for open-ended scientific problem discovery